In this paper, we explore self-supervised audio-visual models that learn from instructional videos. Prior work has shown that these models can relate spoken words and sounds to visual content after training on a large-scale dataset of videos, but they were only trained and evaluated on videos in English. To learn multilingual audio-visual representations, we propose a cascaded approach that leverages a model trained on English videos and applies it to audio-visual data in other languages, such as Japanese videos. With our cascaded approach, we show an improvement in retrieval performance of nearly 10x compared to training on the Japanese videos solely. We also apply the model trained on English videos to Japanese and Hindi spoken captions of images, achieving state-of-the-art performance.
@article{arxiv.2111.04823,
title = {Cascaded Multilingual Audio-Visual Learning from Videos},
author = {Andrew Rouditchenko and Angie Boggust and David Harwath and Samuel Thomas and Hilde Kuehne and Brian Chen and Rameswar Panda and Rogerio Feris and Brian Kingsbury and Michael Picheny and James Glass},
journal= {arXiv preprint arXiv:2111.04823},
year = {2021}
}
Comments
Presented at Interspeech 2021. This version contains updated results using the YouCook-Japanese dataset